Semiglobalization and International Business Strategy

Author(s): Pankaj Ghemawat

Source: Journal of International Business Studies, Vol. 34, No. 2, Focused Issue: The Future of
Multinational Enterprise: 25 Years Later (Mar., 2003), pp. 138-152
Published by: Palgrave Macmillan Journals
Stable URL: http://www.jstor.org/stable/3557149 .
Accessed: 22/06/2011 12:19

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at .
http://www.jstor.org/action/showPublisher?publisherCode=pal. .

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Palgrave Macmillan Journals is collaborating with JSTOR to digitize, preserve and extend access to Journal of
International Business Studies.

http://www.jstor.org
 X1. Journalof International
BusinessStudies(2003)34, 138-152
ll*~~tr-~ 2003 PalgraveMacmillan
0~?C) Ltd. Allrightsreserved0047-2506 $25.00
www.jibs.net

and international business

Semiglobalization
strategy

PankajGhemawat Abstract
If markets were either completely isolated by or integrated across borders,
HarvardBusinessSchool,Boston,MA,USA there would be little room for international business strategy to have content
distinctive from 'mainstream' strategy. But a review of the economic evidence
Correspondence: about the international integration of markets indicates that we fall in between
Dr P Ghemawat, Harvard Business School,
these extremes, into a state of incomplete cross-border integration that I refer
Soldiers Field Road, Morgan Hall 227,
to as semiglobalization. More specifically, most measures of market integration
Boston, MA 02163, USA.
Tel: +1-617-495-6275; have scaled new heights in the last few decades, but still fall far short of
Fax: +1-617-495-0355; economic theory's ideal of perfect integration. The diagnosis of semiglobaliza-
E-mail: pghemawat@hbs.edu tion does more than just supply a relatively stable frame of reference for
thinking about the environment of cross-border operations. It also calls
attention to the critical role of location-specificity in the prospects of distinctive
content for international business strategy relative to mainstream business and
corporate strategy. In addition, it flags factors/products subject to location-
specificity as being salient from the perspective of international business.
Finally, it highlights the scope for strategies that strive to capitalize on the
(large) residual barriersto cross-border integration, as well as those that simply
try to cope with them.
journal of International Business Studies (2003) 34, 138-152. doi:10.1057/
palgrave.jibs.840001 3

Keywords: semiglobalization;globalizationmarketintegrations;marketinperfections;
firmstrategy
location/location-specificity;

Introduction
The first of the three postulates on which Buckley and Casson
(1976, 32) based their theory of the multinational enterprise was
that 'firms maximize profit in a world of imperfect markets.' This
structural insight has proved as fruitful in international business
strategy as it has in 'mainstream' (single-country) business strategy,
where it has been in circulation for even longer. What is somewhat
odd, however, is that work in this vein in international business
strategy has tended to focus on the same sources of market
imperfections as mainstream business strategy: small numbers and,
often related, the business/usage-specificityof key activities,
resources, competencies, capabilities, knowledge, etc., or their
firm-specificityin the sense of being collectively held by the firm's
managerial hierarchy or employee pool and inalienable from it.
However, the obvious potential source of market imperfections
added by the international dimension - the possibly limited cross-
Received: September 2001
Revised: June 2002
border integration of markets or, more generally, the possible
Accepted: August 2002 location-specificityof key activities, resources, etc. - has received less
Online publication date: 27 March 2003 attention. Location-specificity of the specific sort wrought by
 Semiglobalization and international business strateqy
market segmentation at national boundaries is at
the core of this paper.1
This paper consists of two halves. The first half
contains a broad - and therefore inevitably com-
pressed - review of the empirical evidence on the
cross-border integration of markets of different
types: for products (via both trade and FDI), capital,
labor, and knowledge. The review points to the
conclusions that, on the one hand, the observed
levels of cross-border integration of these types of
markets are significant and in many cases have
recently reached highs without historical prece-
dent, but that, on the other hand, the observed
levels of cross-border integration are also very far
from complete and, extrapolating from historical
rates of increase (not to mention recent setbacks),
are likely to remain that way for a long time. This
condition of incomplete cross-border integration,
referred to here as semiglobalization, is more com-
plex than the extremes of total insulation and total
integration because it involves situations in which
the barriers to market integration at borders are
high, but not high enough to insulate countries
completely from each other. Another way of putting
this is that semiglobalization covers the range -
apparently broad as well as complex - of situations
in which neither the barriersnor the links among
markets in different countries can be neglected.
The second half of this paper can be read as a
short essay on the implications of the empirical
finding of semi-globalization for international
business strategy. It begins by noting that semi-
globlization is a sufficient condition for location-
specificity to matter. Although complete market
insulation also suffices, it is a less challenging
condition since, under it, international business
strategy could simply be chunked up into applica-
tions of mainstream (that is, single-location) strat-
egy, performed location by location - although
some problems of coordination would still remain.
Thus semi-globlalization is the underlying structur-
al condition most conducive to thinking in careful
ways about competing across multiple locations
and how that might differ from competing at a
single location. The essay elaborates on this and
other, more specific, implications of the general
diagnosis of semi-globalization. It discusses the
balance to be struck in international business
strategy between attention to location-specificity
and other types of-specificity, and examines the
conditions under which imperfections in particular
types of market (especially knowledge, which was
emphasized by Buckley and Casson, 1976) should
PankajI Ghemawat
139
be granted elevated status. Finally, the essay high-
lights the scope for strategies that strive to capita-
lize on the (large) residual barriers to cross-border
integration, as well as those that simply try to cope
with such barriers.The treatment is meant as much
to stimulate and direct further research as to
summarize research efforts to date.
It is worth adding that the first half of this paper -
the next two sections - focuses on reviewing the
economic evidence about the cross-borderintegra-
tion of markets of different types. The economic
perspective is adopted because economics offers
both a relatively well-developed conceptual frame-
work for the analysis of market integration and
some empirical basis for making judgments about
levels of and changes in cross-borderintegration of
the kinds that occupy its attention. Thus the next
section of this paper looks at the cross-border
integration of product markets, and the section that
follows at markets for various types of resource or
factor - capital, labor, and knowledge. The questions
asked about each type of market concern changes in
its level of international integration, measured in
terms of quantity and price outcomes, over recent
decades or the course of the 20th century, as well as
its absolute level of international integration at the
millennium. Fora more specific delineation of what
is included in and excluded from the review, see
Table 1. While there is arguably a logic to the
pattern of inclusions and exclusions, the more
fundamental point is simply that one cannot talk
about everything in Table 1 in a paper of this scope.
Product market integration
This section begins by looking at the most obvious
quantity measure of the cross-borderintegration of
Table I Dimensions of integration
Dimension Possible
emphases
Criteria for evaluating _ Non-economic
integration
Key boundaries Others
Continents/regions
Localities
Locus of integration Others
Firms
Networks
Type of markets .S .
Input/output emphasis 0 . - D
Outcome variables -
:.S -' ..
Darkshading = primaryemphasis. Gray shading = secondary emphasis.
journal of International Business Studies
* 14-
ra0l
JC|0|lVVIIIII
SOMinlnhAI;7Atl;nn
IIILIUII
1
angi
U inti
IlIL
rnatnnanl
IILIIIIUonal
hime;nipec
C
LI
etr:tionv
z ILCy, PAnleAi
j illll
r.homAwAlt

140

18 for the ratio of merchandise trade to total GDP are

16 - increases for only six of the 11 countries, a median
14 -
change of +2 percentage points from an initial
12 - median value of 20%, and total unweighted
10- increases less than one-half as large as total
8-
,-??i
unweighted decreases.
6' One interpretation of the historical patterns is
4- that:
2
. . . . . (1) trade had taken off in many commodities by the
1820 1840 1860 1880 1900 1920 1940 1960 1980 beginning of the 20th century;
Year
Source: 1820-1992: Angus Maddison, Monitoringthe WorldEconomy 1820-1992, OECD 1995 1993-1998: (2) there were substantial increases in the trade of
World Trade Organization and InternationalMonetary Fund data.
manufactures over the course of the 20th
Figure 1 Exports divided by GDP. Source: 1820-1992, Maddi- century, particularly its second half; and
son (1995); 1993-1998, World Trade Organization and Inter- (3) the service sector continues to be a very large
national Monetary Fund data. bottleneck for trade-relatedflows even though it
is growing.
product markets: trade flows. It then looks at
Irwin's (1996) comparison of the composition of
foreign direct investment (FDI) stocks and, finally
and very briefly, at cross-borderprice integration. US merchandise trade over a century is suggestive
in this regard:see Table 2. While this neat ordering
of the globalization of commodities, manufactures
Trade flows and services is obviously an oversimplification, it is
To begin with a very long-run perspective, consider nevertheless useful.
data on world exports divided by world GDP (the So trade has clearly increased over the last 50, 100
usual normalization) over the last two centuries and 200 years. But it is useful to supplement this
based on and updated from data in Maddison observation with some data about the absolute
(1995). As Figure 1 indicates, this ratio increased level of integration of product markets through
from about 1%at the beginning of the 19th century trade. Economists who study international trade
to nearly 10% towards the beginning of the 20th
generally do not regard trade intensity as very high
century, and, despite a period of stagnation and in absolute terms. In fact, they tend to find the
decline bounded by the two World Wars, has since issue of why there is not much more trade more
managed to edge up towards 20%. Trade intensity interesting than the new records being set. To see
has clearly reached new heights in the last quarter the room for increase, consider a hypothetical
of the 20th century.
benchmark, suggested by Frankel (2001), in which
The increase in trade intensity over the course of national borders did not affect buying patterns at
the 20th century looks all the more remarkable all. In such a situation, buyers in a particularnation
when one accounts for the increasing share of GDP
contributed, especially in developed countries, by
two sectors that account for relatively little trade - Table 2 Commodity composition of US merchandise trade
services and government. One way of stripping out
Year Percentage distribution
the effects of these 'non-traded' sectors is to remove
them from the calculations and focus on the ratio Exports Imports
of merchandise trade to merchandise value added.
This leads to striking increases in measured trade Agricultural goods
1890 42.2 33.1
exposure, as illustrated by Feenstra's(1998) sample 1990 11.5 5.6
of 11 relatively developed countries between 1913
Raw materials
and 1990. Over this period, the ratio of merchan- 1890 36.6 22.8
dise trade to merchandise value added increased for 1990 11.6 14.8
nine of these countries; the median change was Manufactures
+22 percentage points, compared with an initial 1890 21.2 44.1
median value of 36%, and total unweighted 1990 77.0 79.6
increases were close to 20 times as large as total Figures may not total to 100 due to rounding. Agriculturalgoods
unweighted decreases. The corresponding statistics includes processed foods. Source:Irwin(1996).

Journalof InternationalBusinessStudies
 Semiqlobalization and international business strateqy PankajI Ghemawat
141
141

100
95-
border and language (mostly) and have friendly
90- relations with each other, making theirs the
85
80- largest bilateral trading relationship in the world
(McCallum, 1995). The free trade agreement signed
-_ 75-
I
70-
* BEL
;
C
65-
60-
I in 1988 between the two countries did reduce this
C 55-
. I
domestic multiple by the mid-1990s, but only to 12
50- NET

4 45- (and with the multiple remaining stuck at 30-40 in

40-

MEX
*CAN
WSWE
the case of services) (Helliwell, 1998, Chapter 2).
SWI - --: KOR
JK
Cruderdata suggest a multiple of about six for trade
: GER
within as opposed to between the member states of
SP
RUS . o. i.i..
.. FR
AUS :::: CHi IT
10-
ARG
US
I
the European Union (Helliwell, 1998, Chapter 3).
5

0
BR
IBR
I
JAFp

I I I I I Given the regionalization of world trade that has

0 10 20 30 40
Share of World GDP (%)
Sources:GNP Rankingsbased World Development Bank'sWorld DevelopmentIndicators.
2000.Import/ExportData from WTO.
Figure 2 Actual vs perfect product market integration through
trade.
would be as prone to obtain goods and services
from foreign producers as from domestic ones, and
the share of imports in total domestic consumption
would equal 1 minus the nation's share of world
product. For example, as the US economy accounts
for about one-quarter of gross world product, the
US import/GDP ratio would, at this benchmark,
equal 1 minus the US share of world production, or
0.75, as would, under the first-orderassumption of
balanced trade, the US export/GDP ratio. However,
the actual ratios are only about one-sixth as large as
these hypothetical levels!2
The line with slope -1 in Figure 2 traces out this
hypothetical benchmark of perfect product market
integration as national shares of world product
vary. It also plots the position of the 20 largest
nations in these terms. Notice that most of the
nations cluster close to the origin, and all fall well
below the hypothetical maximum - including the
two high-fliers, Belgium and the Netherlands.
While the hypothetical benchmark suggests sig-
nificant barriers to cross-border product flows, it
also embodies a number of extreme assumptions. A
real example that points in the same direction is
provided by Canadian provinces' patterns of trade
with each other compared with their trade with the
USA. In addition to the fact that data for these
patterns are available, they have the added advan-
tage of involving (international) trading partners
that are close to each other along a number of
dimensions. As of 1988, trade linkages between
Canadian provinces were 20 times as large as their
linkages with the 30 US states that traded the most
intensively with Canada. This was true despite the
fact that Canada and the USAshare a common land
50 60
70 80 90 100
been under way, the multiples of domestic-to-
international economic exchange would obviously
be higher if one were comparing trade within
countries with trade outside the regional blocs to
which they belong.
To sum up, trade intensity has clearly reached
unprecedented levels, but still reveals significant
impediments to the cross-border integration of
product markets.
Foreign direct investment
Tradeis not the only way in which the cross-border
integration of product markets might be accom-
plished: FDI, which involves product-specific
investment across borders, is an obvious alterna-
tive. To start with a long-run perspective, consider
data on FDI stocks divided by GDP over the last
century based on calculations in WorldInvestment
Reportsissued by the UN Center on Transnational
Corporations. As Table 3 indicates, FDIsurvived the
interwar years better than trade (it even came to
substitute for the latter as tariff barriers rose), but
did not take off again quite as rapidly in the
immediate postwar years. FDI has surged, however,
since 1980 and, by 1997, had come to exceed the
previous (prewar) peak in its share of gross world
GDP by a significant margin: 12% to 9%. Despite
the declines in the ratio of outward FDI stock to
GDP exhibited by the UK and France, the largest
foreign investors prior to World War I, the aggre-
gate comparison is suggestive of an increase to
unprecedented levels. In sectoral terms, FDI has
mirrored trade over this time period by shifting
away from natural resources and raw materials (the
'primary'sector) towards manufacturing and, more
recently, services.
Obviously, such historical comparisons come
with some caveats. For one thing, they are affected
in important ways by fundamental shifts in relative
exchange rates (and purchasing power). For
another, they are based on book values rather than

Journal of International Business Studies

 Semialobalization and international business strateav .PankaiGhemawat
......
*
142 .142. .

Table 3 Outward FDIstock as a percentage of GDP

1914 1938 1960 1980 1985 1990 1995 1997

France 21.1 27.8 6.8 2.7 6.0 9.2 12.0 13.6

Germany 11.1 0.8 1.1 5.3 9.7 9.2 11.1 14.4
Japan 0.8 9.9 1.2 1.9 3.3 6.9 4.7 6.5
UK 52.3 38.5 15.0 15.0 21.9 23.8 28.3 29.1
USA 7.2 8.5 6.2 8.1 6.2 7.9 10.0 10.6
World 9.0a 4.4 4.8 6.4 8.5b 11.8
a1913 data. b1991 data.
Figurefor 191 3 is an estimate. Sources:1913-1991, World Investment Report 1994; 1997, World Investment Report 1999.

on market values of FDI. The magnitude of this 100.0-

omission seems to be large: data compiled by the 90.0'

US Commerce Department suggest that measure- 80.0

ment on the basis of market values rather than 70.0

book values doubles the estimated values of both -

60.0
US FDI abroad and FDI in the USA. One could argue >, 50.0
SWE
a, " NET
that this omission leads to greater underestimation (5
.E
C~ 40.0' BEL
of the true values of FDI stocks towards the end of u.
30.0 UK
the 20th century than towards its beginning, BR
because of higher inflation rates (until relatively MEX 20.0
ARG
xCAN
- "
CH
AUS
RUS
KOR
swI 10.0 +I FR
recently) in the modern period and the increased SP- IN; GER JAP & US
I'l
importance of intangible assets that are more prone u.u
0.0
I
10.0
',
20.0 30.0
...
40.0
.
50.0
.
60.0
.
70.0
.
80.0 90.0
-

to slip through accountants' nets.
Once again, it is useful to look at the current level
of integration of product markets through this
channel in absolute terms, not just in relation to
the levels experienced earlier. Assume, as in the
analogous calculation undertaken earlier for trade,
that inflows/outflows are, to a first approximation,
balanced, and consider a country that accounts for
x% of world investment. Then, if national borders
did not affect investment patterns at all, foreign
capital would account for (100-x)% of total
investment in that country. The line with slope
-1 in Figure 3 traces out this hypothetical bench-
mark of perfect integration as a function of national
shares of gross fixed investment (x). It also plots the
position of the 20 largest nations in these terms,
based on their recorded FDI inflows. As in the case
of trade, most of the nations cluster close to the
origin, and all fall well below the hypothetical
maximum. Also note that this broad conclusion
would not be affected by looking at FDI outflows,
although the positions of individual countries
would shift substantially. China, for instance,
would be less of a high-flier.
Overall, FDI intensity has, like trade intensity,
reached unprecedented levels while continuing to
fall far short of the levels that would be implied by
100.0
Share of World GFI
Source: EIU Country Data
Figure 3 Actual vs perfect product market integration through
FDI. Source: EIU country data.
perfect cross-border integration of product markets
through this channel.
Price integration
Viewed in terms of prices rather than quantities,
the ultimate in market integration is achieved
when two (or more) markets are yoked together
by the so-called law of one price (LOP) - that is,
prices equalize across them. Implicit in LOP is a
(strong) zero-arbitrage-profits principle. Note that
the degree of price integration of product markets
can be high even when the quantity flows across
them are limited - for example, for some commod-
ities whose local prices are pegged to world
benchmark prices, including ones with high value-
to-weight ratios. As a result, economists often treat
tests of market integration based on prices as being
more definitive than tests based on quantities.
Quantity-based tests of cross-border market inte-
gration predominate, nonetheless, because, except
for (nearly) perfect commodities, tests of price
integration are generally hampered by the lack of

Journalof InternationalBusinessStudies
 Semiqlobalization and international business strateqy PankajGhemawat
143

data on local currency prices of identical products
across countries. The relatively few studies of
products and services that meet these objections
generally indicate substantial, sustained departures
from LOP.Cross-country price dispersions tend to
be large and to die down at a slow pace, and there is
little evidence of recent movement toward smaller
dispersions or speedier dampening (Rogoff, 1996).
In conjunction with the data presented earlier
concerning integration through trade and FDI
flows, an overall inference that product market
integration has increased significantly in recent
decades, while continuing to fall far short of
perfection, seems most plausible.
Factor market integration
Product markets are not the only type of market
whose cross-border integration one might find
interesting; factor markets of various types are also
candidates for attention. This section presents and
discusses evidence on the extent of cross-border
integration of markets for capital, labor, and
knowledge, in that order. Both quantity-based and
price-based measures of integration are looked at
wherever possible.
Capital
The previous section's discussion of FDI can be
broadened to look at international capital flows
over the last 100 years.3 Because of identities in
national income accounting, countries' net capital
flows can be measured as the reverse of their
current account balances. Data assembled by
Obstfeld and Taylor (1997) on absolute net capital
flows divided by GDPs for 12 countries suggest that
this index of capital mobility has increased in
recent decades, but was higher still around the
beginning of the 20th century (see Table 4). Note
that the impressive performance 100 years ago was
accomplished in spite of informational and con-
tracting problems. Such problems were, most likely,
much more severe given the lack of generally
accepted accounting principles and commensu-
rately weak reporting requirements.
Of course, not all capital flows are equally
important from the perspective of economic globa-
lization. In particular, the recent period has seen a
surge in short-run flows, or at least transactions,
that is most strikingly evident in the volume of
foreign exchange transactions, which exceeds $1
trillion daily. Foreign exchange trading can, how-
ever, be regarded as a response to a source of
volatility - exchange rate risk - that was mitigated
significantly in the earlier period by the prevalence
of the gold standard. For this reason, and because
most trades of this sort seem to be purely spec-
ulative, it is problematic to use the size of foreign
exchange markets today to infer a much greater
level of cross-border integration of capital markets
than at the beginning of the century.
This suggestsfocusing attention on long-run capital
flows, which include portfolio investment as well as
FDI. Portfolio investment has increased significantly
in absoluteterms in recent decades,but seems to have
failed to keep pace with FDI, with its share slipping
from about two-thirds of total long-run cross-border
investment in the early 20th century to about one-
half today (Bloomfield,1968). Nevertheless,the range
of securities traded today across borders is much
broader,in type as well as in number - a shift that,
some argue,has contributedto increasedcross-border
integration along this dimension.

Table 4 Size of net capital flows since 1870 (mean absolute value of currentaccount as percentage of GDP, annual data)
Period Arg Aus Can Den Fra Ger Ita lap Nor Swe UK USA All

1870-1889 18.7 8.2 7.0 1.9 2.4 1.7 1.2 0.6 1.6 3.2 4.6 0.7 3.7
1890-1913 6.2 4.1 7.0 2.9 1.3 1.5 1.8 2.4 4.2 2.3 4.6 1.0 3.3
1914-1918 2.7 3.4 3.6 5.1 11.6 6.8 3.8 6.5 3.1 4.1 5.1 a
1919-1926 4.9 4.2 2.5 1.2 2.8 2.4 4.2 2.1 4.9 2.0 2.7 1.7 3.1
1927-1931 3.7 5.9 2.7 0.7 1.4 2.0 1.5 0.6 2.0 1.8 1.9 0.7 2.1
1932-1939 1.6 1.7 2.6 0.8 1.0 0.6 0.7 1.0 1.1 1.5 1.1 0.4 1.2
1940-1946 4.8 3.5 3.3 2.3 3.4 1.0 4.9 2.0 7.2 1.1 3.2a
1947-1959 2.3 3.4 2.3 1.4 1.5 2.0 1.4 1.3 3.1 1.1 1.2 0.6 1.8
1960-1973 1.0 2.3 1.2 1.9 0.6 1.0 2.1 1.0 2.4 0.7 0.8 0.5 1.3
1974-1989 1.9 3.6 1.7 3.2 0.8 2.1 1.3 1.8 5.2 1.5 1.5 1.4 2.2
1989-1996 2.0 4.5 4.0 1.8 0.7 2.7 1.6 2.1 2.9 2.0 2.6 1.2 2.3
Source:ObstfeldandTaylor(1997).
a: Average with some countries missing.

Journal of International Business Studies

 Semiglobalization and international
144
International financial crises represent the flip side
of international capital mobility. Once again,
historical comparisons suggest that international
financial crises, particularly in emerging markets,
are not without precedent. Thus data on the
currency and banking crises experienced by 21
countries between 1880 and 1998 indicate that the
most severe crises, on average, were in the interwar
period, followed by the prewar period; postwar
crises, in contrast, have been milder in terms of the
drops in output experienced, and shorter-lived
(Bordo et al., 1999). And even when the sample is
restricted to emerging countries, recent levels of
instability do no worse than 'match' prewar levels,
in which the gold standard acted as a crisis
transmission belt, and emerging countries, at least,
tended to lack lenders of last resort.4
In addition to these historical comparisons,
quantity-based measures also permit some infer-
ences about the absolute level of cross-border
integration of capital markets. As in the case of
trade, the professional curiosity of economists has
focused on smaller-than-expected flows (or stocks).
Probably the most famous 'anomaly' of this sort is
the one uncovered by Feldstein and Horioka (1980),
who calculated a 90% correlation between domes-
tic savings and domestic investment across a panel
of countries. Their estimate is much higher than
benchmark models that assume perfect capital
mobility would lead us to expect. Another anomaly
that points in the same direction concerns what is
called home-country bias: investors in each country
hold much largerproportions of their wealth in the
form of domestic securities than they would with
internationally well-diversified portfolios. Thus, by
one estimate, US investors should have held more
than half their wealth in foreign equities in the
1980s, instead of the less than 10% that they
actually held (Lewis, 1995).
Price-based measures of capital market integra-
tion - with price integration reinterpreted in terms
of the equalization of rates of return on common or
comparable securities across national boundaries -
supply additional evidence about the continued
segmentation of capital markets. One benchmark
example is provided by Obstfeld and Taylor's(1997)
comparison of 1-year interest rates on sterling-
denominated assets sold in London and in
New York over the last 100-plus years. Figure 4
tracks the standard deviation of differences in
returns in the two cities as an inverse measure of
capital market integration. The data indicate sig-
nificant cross-borderintegration of capital markets
journal of International Business Studies
business strateqy PankajI Ghemawat
prior to 1914, the breakdown of that integration in
the interwar period, and its slow restoration in the
postwar period. Qualitatively similar conclusions
are suggested by comparing real rather than
nominal returns, although that does increase the
standard deviation of the dispersion of returns,
presumably reflecting the effects of currency risk,
both nominal and real.5 At a more macro level,
studies of returns, such as Bekaert and Harvey
(1995), indicate that the cointegration of capital
markets varies greatly in its level and extent over
time.
Overall, like product market integration, capital
market integration has increased significantly in
recent decades, but seems to continue to fall far
short of perfection.
Labor
Data on the cross-border integration of labor
markets are sparser than for product or capital
markets. However, they generally suggest that the
number of international migrants (defined as
people residing in foreign countries for more than
1 year) has grown with world population in recent
decades, but represents a smaller share of world
population than 100 years ago. With regard to the
first point, there were, according to the World
Migration Report, an estimated 150 million long-
term international migrants in 2000, or 2.5% of
world population (Martin, 2000). The comparable
numbers for 1965 were 75 million migrants and
2.2% of world population.
Over a longer time frame, the period between
1880 and 1915/1920 stands out as the heyday of
international migration. During these years, 32
million people migrated from Europe, most of
them to the USA (Kenwood and Lougheed, 1989).
In addition, there were 6-8 million net migrants -
mostly 'coolie' or indentured labor - from India,
3.50
3.00-
2.50 -
2.00 -
1.50
1.00-
0.50 -
II IIII
1880 1902 1916 1922 1928 1935 1943 1952 1962 1974 1985 1993
Source: MauriceObstfeldand AlanTaylor,"TheGreatDepressionAs A Watershed:International
CapitalMobilityOverthe LongRun,"NBERWorkingPaper 5960, March1997.
Figure 4 Standard deviation of nominal return differentials.
Source: Obstfeld and Taylor (1997).
 Semiglobalization and international
China, and other Asian countries to the rest of the
world (Held et al., 1999, 293-295, 311). Adding in
other cross-bordermovements could push the total
past 45 million, or 3% of world population in 1900.
Higher migration rates 100 years ago are also evident
in country-level data - for example, for the largest
receiver,the USA.Thus census data indicate that 14%
of the US population was foreign-bornat the turn of
the century,comparedwith 10%today (Dune, 2001).
Note that, through a substantial part of the earlier
period, a number of large receivers, including the
USA,placed no restrictionson immigration.
Turning from quantity-based to price-based mea-
sures, the most obvious indicator of cross-border
integration of labor markets would be the cross-
border convergence of wages. Data on the evolu-
tion of average per capita incomes (a rough and
ready proxy for average wages) indicate that, while
incomes in industrialized countries have tended to
converge over the last few decades, a few Asian
'tigers' have been the only countries able to break
away from the rest of the developing world and
catch up with the industrialized world (see
Figure 5).6 More sophisticated
conclusion, and indicate that the failure of most
developing countries to catch up can be reconciled
only with a weaker notion of convergence -
conditional convergence (Barro and Sala-i-Martin,
1995). Conditional convergence allows for differ-
ences in the steady-state incomes toward which
different economies are trending, based on differ-
ences along dimensions such as investment, educa-
tion, and population growth. Human capital turns
out, in attempts to fit conditional convergence
models to the data, to have a particularly marked
effect on the predicted extent of convergence.
sO18,000
$16.000
S14.000
S.o00))
$6.0()0 '-
.48z. . '
$2.000
$()
1950() 1953 1956 1959 1962 1965 196X 1971 1974 1977 1980 1983 1986 1989 1992 1995
Source: Bruce Scott, Economic Strategies of Nations, adapted from the
PennWorldTablesandthe WorldBank.
Figure 5 Convergence? GDP per capita across economic
groups, 1950-1997 (PPP-adjusted). Source: Scott (2000),
adapted from the Penn WorldTablesand the World Bank).
business strategy Pankaj Ghemawat
145
Taking a somewhat longer view, it is worth
emphasizing that the 19th century apparently saw
a divergence, rather than a convergence, of
incomes across countries that has been only
partially reversed in the 20th century (Baldwin
and Martin, 1999). So, over that kind of time frame,
the dispersion of incomes across countries
increased, in net terms, instead of decreasing. This,
along with the other data presented in this
subsection, would seem to imply skepticism about
the extent to which labor markets have integrated
across national boundaries.
Knowledge
The other types of cross-border flows that have
been discussed already can carry knowledge across
national borders as well, since it can be congealed
in products, embedded in capital equipment,
vested in skilled personnel, etc. Given the topics
already covered in this paper, this subsection will
focus on cross-border flows of knowledge in pure,
disembodied form. In addition to rounding out the
coverage, this focus has the advantage of offering a
tests confirm this relatively simple benchmark: as disembodied
knowledge has a 'non-rival' character - that is, as
its use in one market, whether defined in geo-
graphic or product-related terms, should not pre-
clude its application to others - perfect cross-border
integration in this context should imply that
knowledge, once developed anywhere in the world,
is available everywhere else as well.
The conceptual simplicity of focusing on disem-
bodied knowledge flows does, however, exact an
empirical toll: because of their intrinsic intangibil-
ity, such flows are particularlyhard to measure. The
evidence presented in this subsection is corre-
spondingly sketchy. It tentatively suggests, how-
ever, that there have been substantial increases in
In(tlustrialized Countries
cross-border knowledge flows over time and, a bit
,. Asian lTigers*
more definitely, that cross-border integration in
:' :: inciludiesHonLg
.
Kollng.
Sollth
Silngapore
Korlca. and
this regard nevertheless remains very incomplete.
Consider these inferences in turn.
With regard to technological knowledge, cross-
border licensing provides one indicator that
supports the inference of increased cross-border
.......... Other D)eveloping
:". - COUl trieS
Sub-Saharlan Afrl-ica knowledge flows over time. Such licensing is not
India
new - international royalties accounted for a
significant component of James Watt's receipts
from his steam engine patents in the early nine-
teenth century, for example. However, the available
data, along with informational and contracting
problems that were even more acute early on than
they are now, suggest that the voluntary transfer of
Journal of International Business Studies
* - Vam-SlAl*hAi17tatinn
sIIIILuIa/iV all/ll
anti ;IIILI
aU
-ntornat;innal
n IILI,11
146
knowledge across national borders is far more
common than it used to be. Concerning more
general managerial knowledge, the post-World War
II period, in particular,has seen the development of
new types of organizations and organizational
forms that have also facilitated knowledge transfer.
Franchising, which really emerged in its modern
form in the USA in the 1950s, is one example. And
management consulting firms, which began their
international expansion at roughly the same time,
are regardedas having evolved into major channels
for the international diffusion of new managerial
techniques (Micklethwait and Wooldridge, 2000).
Of course, the spread of multinational enterprises,
intent on applying the same technological and
managerial knowledge to more and more markets,
points in the same direction. So, arguably,does the
explosion in cross-borderinformation transmission
capacity since the early 1980s.
These increases in cross-border knowledge flows
notwithstanding, there are also numerous indica-
tions of the continued geographical localization of
knowledge. The survey evidence on the size of
knowledge transfer costs, although not altogether
satisfying, is suggestive. An influential study by
Teece (1977) concluded that transfer costs
accounted for an average of 19% of total project
costs - and ranged from 2 to 59% - in a sample of
technology transfers in the chemicals, petroleum
refining, and machinery sectors. Outcome-based
perspectives that point in the same direction are
numerous. Through the 1980s, nearly 90% of the
US patents taken out by the world's 600 largest
corporations listed the inventor as a resident of the
corporation's 'home base' (Patel and Pavitt, 1994).
Patents whose inventors reside in the same country
are typically 30-80% more likely to cite each other
than inventors from other countries, and, on
average, these citations come 1 year sooner (Jaffe
and Trajtenberg,1999). A recent study of R&D and
productivity spillovers across large OECD econo-
mies estimated the average elasticity of such spil-
lovers with respect to distance as -1 to -2.4%
Table 5 Strategydomains
Focus
Increasingattention to
location-specificity/non-specificity
Journalof InternationalBusinessStudies
hriMAinpc
oI/IIIsN,
rrsatAnv
L ILtqy Pknie;i (C.h,-mAvAt
ra iKaj 1IncmlVawat
(Keller, 2000). The importance of locally dense
information flows is also evident in inter-
nationally successful geographic clusters.7 Such
perspectives remind us that, although the avail-
ability of information transmission capacity may
help knowledge to travel across national borders, it
is far from sufficient to make knowledge perfectly
portable.
Semiglobalization as a research program
In summary, most measures of cross-border eco-
nomic integration have increased significantly in
the last few decades, but still fall far short of the
theoretical extreme of total integration. This
empirical conclusion of semiglobalization is valu-
able in and of itself given the ongoing debate
between two polar perspectives: one maintaining
that we have achieved a state of (near) globality, in
which there is so much integration across national
borders that the latter can, for many practical
purposes, be ignored, and the other professing
skepticism that there is anything fundamentally
new about the levels of cross-border integration
that have been achieved to date (Giddens, 1996;
Held et al., 1999). It seems possible to achieve some
closure to this debate, at least in the economic
arena.
As a bonus, semiglobalization affords - unlike
alternate possibilities - room for international
business strategy to have content that is distinctive
from 'mainstream' (single country or location)
business strategy or, for that matter, corporate
strategy.To make this point as precisely as possible,
it is useful to classify the field of strategy into the
domains depicted in Table 5. Note the somewhat
paradoxical character of domain 1, mainstream
business strategy: by assuming total specificity, it
allots the least attention to understanding either
business/usage-specificity or location-specificity. As
a result, we have to look to domain 2, that of
mainstream corporate strategy, for interesting ana-
lyses of variations in the extent to which key firm
activities, resources or knowledge are business-
Increasingattention to business-specificity/non-specificity
Single business Multiplebusinesses
Single country/location 1. (Mainstream) 2. (Mainstream)
business strategy corporate strategy
Multiple countries/locations 3. International 4. International
business strategy corporate strategy
 CSmPMnlnhall7At;nn anti ;ntornat;^nn:l
n an1U
acmIIIIY.iiVz.LIoV IIICIIIIIIl
specific as opposed to generic (fungible across
businesses). And we must also look to domain 3,
that of international business strategy, for analyses
of variations in the extent to which activities,
resources or knowledge are location-specific as
opposed to free-flowing (fungible across locations).
Domain 4, featuring international corporate strat-
egy, purports to combine both business/usage-
specificity and domain-specificity, but it is the one
about which we currently know the least.
The key point to be made here is that semigloba-
lization and the location-specificity or geographic
segmentation of markets implicit in it is critical to
the possibility of domain 3 having content qualita-
tively distinct from domains 1 and 2. Begin by
comparing domains 3 and 1. The critical role of
semiglobalization can be illustratedby contrasting it
with the extreme alternatives of markets totally
insulated from each other by national boundaries or,
at the opposite extreme, perfectly integrated with
each other across them. Obviously, with complete
market insulation, firms could simply decompose
their choice problems into country-sized chunks.
And if marketswere completely integrated with each
other, the analysis of multiple countries could, once
again, be folded back to the single-country base case
that is the staple of mainstream business strategy
(domain 1), as there would effectively be a single
large country. Situations with intermediate levels of
cross-borderintegration cannot be dealt with in the
same way, however, in that they do not lend
themselves to purely country-level analysis.8
Next, compare domain 3 with domain 2. The role
of semiglobalization or, more precisely, location-
specificity in affording scope for international
business strategy to have content distinctive from
mainstream corporate strategy is, perhaps, subtler
but no less important than in the previous case.
Specifically, note that the insights into firm bound-
aries and expansion derived, respectively, from
Coase (1937) and Penrose (1959), were not only
worked into international business strategy by
Buckley and Casson (1976), among others, but also
into mainstream corporate strategy by, in particu-
lar, a large body of work on corporate diversifica-
tion. So, although such insights have been very
valuable, they do not by themselves imply content
for domain 3 that is conceptually or otherwise
qualitatively distinct from that of domain 2; they
are a common element of both. For that, what is
needed is attention to operations across multiple
locations that are distinct from, but not entirely
independent of, each other.
h..eZirs s+wa+vxv D:-ni- r.--m^lt
UUslllns. slraLeqy rallIdJ unllItIlaWdL
147
Looking more broadly across domains 1-3, Table5
indicates that location-specificity must be invoked
to distinguish domain 3 from mainstream strategy
of the business and corporate varieties (domains 1
and 2). Semiglobalization ensures such location-
specificity, and therefore supplies a conceptually
coherent foundation for further analyses at the
market and firm levels.
Market/factor-level issues
The preceding argument is equivalent, in some
respects, to saying that international business
strategy should pay more attention to market
imperfections involving location-specificity rather
than business/usage-specificity. Those who work
primarily on the latter are likely to be somewhat
skeptical. One frequently cited concern in this
context is the argument that business/usage-speci-
ficity affords more room for firm-specific advan-
tages (and disadvantages) than location-specificity.
But given complementarities among activities,
resources, etc., this argument is a bit of a red
herring.
To see why, consider a stylized example in which
there are two factors - knowledge, denoted by N (to
avoid confusion with K for capital), and labor,
denoted by L, with N entirely business/usage-
specific and subject to internalization pressures as
a result, and L entirely location-specific. Given
complementarities between L and N, profit-max-
imizing firms cannot afford to ignore the labor cost
differences across their various cross-borderoptions
even if their management of L itself does not offer
the prospect for sustainable firm-specific advan-
tages. In particular,if cross-borderdifferences in the
cost of L loom sufficiently large, economic viability
will require either that they be capitalized on or
that some powerful way of countering them be
found. It is hard to see how creative thinking along
either of these lines is fostered by suppressing
consideration of location-specificity, even if it
applies only to a 'generic' factor, L. And even if
labor-cost variations cannot underpin sustained
competitive advantages for the firms that exploit
them because all competitors tap into them,
exploitation of them may be necessary to avoid
unsustainable disadvantages. The whole point of
incomplete integration, after all, is that such factor
price equalization will occur, if at all, only in the
very long run and cannot, therefore, be assumed in
decisions being made in the short-to-medium run.
Analogous points can be made in the context of K
as opposed to L. Note that if capital markets were
Journal of International Business Studies
 Semialobalization and international
--l- -11---
148
148
perfectly integrated, there would be one global pool
of capital available to fund ventures, and decisions
on whether to proceed with investments could be
separated from decisions about how to finance
them. Such separation of investment and financing
decisions, while often assumed domestically, does
not fare well in an international context. Foreign
investment is, to a significant extent, financed
locally in the host country. Thus Feldstein (1995)
concluded that only 20% of the value of assets
owned by US affiliates abroad was financed by
cross-border flows of capital from the USA, with an
additional 18% accounted for by retained earnings
and the rest representing financing with foreign
debt and equity. In such a context, it is hard to
believe that MNEs allocate capital globally to
equalize marginal returns on investment projects
wherever they are undertaken. Instead, firms'
investments in real assets seem to be affected by
local financing possibilities - or wealth effects. And
the impact of financial variables on real ones may
be more than marginal: some major merger and
acquisition waves, for example, seem to have been
driven, in large part, by changes in exchange rates
(e.g. Blonigen, 1997). This is just one of many areas
for additional research related to semiglobalization
- in this case, concerning segmented international
markets for capital and how they interact with real
(non-financial) variables.9
The broader point that emerges from this discus-
sion is that semiglobalization or incomplete inte-
gration is often underplayed because of inadequate
attention to the location-specificity of L and K on
the grounds that they are generic factors of
production incapable of sustaining firm-specific
advantages. Capital also seems to get pulled down,
as markets for it are supposed to be subject to a high
degree of cross-border integration10 and labor
because it is seen to represent a 'low' basis for
cross-border competition. In any case, whatever the
precise reasoning, the effect is to devalue capital
and labor for being relatively non-specialized
factors and to focus attention on knowledge. This
may seem a reasonable approach. However, recall
that it is controverted by the evidence, summarized
in the previous section, that markets for capital and
labor, just like markets for knowledge, exhibit
significant barriers to cross-border integration. As
a result, even the apparently unspecialized factors
of capital and labor are specialized at the level of
location, if in no other sense. Thus they can assume
strategic importance in an international context
and should be attended to.
Journal of International Business Studies
business strateav
- ---=pi
Pankai Ghemawat
. -
--)
....... - . . - I
Having said that K and L merit more attention
than they have historically attracted, it must be
added that this is not necessarily inconsistent with
the focus of much of the relevant literature,
including Buckley and Casson (1976) early on, on
knowledge, or as the key factor underlying the
market imperfections that are most critical for
international business. Instead, what the discussion
implies in this regard is that claims of special status
for N as a factor in international business strategy
(domain 3 in Table 5) have to be based on the
location-specificity of N. Otherwise, international
business strategy and multimarket corporate strat-
egy will be difficult to differentiate. Also note that
in some cross-border contexts, at least, considera-
tions of location-specificity do seem to dominate in
knowledge-related decision making (e.g. Alcacer
and Chung, 2001). Nevertheless, there would seem
to be great demand for additional research on this
much-discussed topic.
Firm-level issues
In addition to flagging factors/products subject to
location-specificity as being salient from the per-
spective of international business strategy, the
diagnosis of semiglobalization sheds some light
on the content of such strategy at the firm (as
opposed to market) level. Most broadly, semigloba-
lization significantly enriches the strategy space
open to firms relative to the straitjacketing struc-
tural extremes of (1) complete isolation at the
borders, which would dictate localization, and (2)
complete integration, which would dictate standar-
dization. Cases intermediate to 'one country' and
'one world' present decision-makers with more
than one obvious strategy option. Therefore these
cases require some higher-level decisions about
how their firms are going to compete to add value.
There are many specific ways in which firms
might try to add value through cross-border opera-
tions under conditions of incomplete integration,
but they can be grouped in terms of two funda-
mental economic functions - in the sense of
mechanisms for adding value, as opposed to
marketing, production, etc. - that organizations
try to fulfill by crossing borders. The first function,
aggregation, involves exploiting the similarities
across countries, while somehow side-stepping the
differences among them, so as to tap increasing
returns to scale. The second, arbitrage, involves
exploiting differences among countries by taking
advantage of variations in absolute costs or will-
ingness-to-pay. The prototypical aggregator is a
 Semiglobalization and international business strateqy
firm that takes advantage of (partly) locationally
mobile resources subject to increasing returns to
perform roughly the same activities in different
countries (a 'horizontal' MNE). The prototypical
arbitrageurtakes advantage of international differ-
ences by geographically separating activities in an
integrated vertical chain (the vertical MNE).
Arbitragewas the function that dominated early
international economic activity, as evident in the
operations of the trading companies chartered in
the 16th and 17th centuries, the whaling fleets of
the 18th century, and the vertically integrated
agriculturaland extractive (mining) companies that
emerged in the 19th century (Ghemawat, 2000). In
contrast, aggregation first came to the fore - with
the possible exception of a few international bank-
ing chains that emerged earlier in the 19th century
- with the manufacturing multinationals that
began to appear in the second half of the 19th
century. Despite this late start, however, casual
evidence suggests that aggregation has comman-
deered researchers'attention to the point where the
arbitrage function is often ignored. The long-
running discussion of the tensions between inte-
gration and responsiveness and their resolution is a
good example (Prahalad and Doz, 1987). These
issues are salient in the context of aggregation, but
not in the context of arbitrage, which is often
passed over in silence as a result. Note that such a
bias towards aggregation would lead to suboptimal
responses to conditions of incomplete integration
because of an undue emphasis on treating impor-
tant differences across countries as sources of
difficulty to be ignored or minimized (as part of
an aggregation approach) rather than as possible
sources of value (as part of an arbitrage approach).
To consider all possible levers of value, it is
important to supplement horizontal approaches
that emphasize aggregation with vertical
approaches that seek to capitalize on (as opposed
to merely cope with) differences - that is, that
emphasize arbitrage.
Unbundling the two functions should help in
this regard. To start with the one that tends to get
overlooked more, arbitrage,the schema used earlier
in this paper to distinguish among markets for
products, capital, labor, and knowledge also sug-
gests a correspondingly broad array of arbitrage-
based mechanisms for (potentially) adding value.
Firms can arbitrage the incomplete integration of
product markets across borders by becoming tra-
ders. Capital market differences provide them with
a strong incentive to account for international
Pankaj Ghemawat
149
differences in the cost of capital. They can arbitrage
labor cost differences by relocating labor-intensive
activities to countries with low labor costs. And
they can try to harness knowledge differences and,
more broadly, geographically dispersed knowledge
by making asset-seeking (rather than asset-exploit-
ing) investments in critical locations - a task that
involves detailed coordination across multiple
locations rather than, as some would have it, the
death of geography.
The aggregation function also lends itself to
unbundling. Here, there are continua of possibili-
ties ranging, as noted above, from the complete
localization of a business by country at one
extreme to complete standardization across coun-
tries at the other. Interestingly distinct - and
progressively less researched - intermediate possi-
bilities include:
(1) adaptation, in which the business model origi-
nated in the 'home base' becomes the basis for
local modification;
(2) platform or front-to-back approaches, in which
certain core features of a business model (the
'platform') are preset globally, while others can
be altered in light of local conditions; and
(3) clustering, which emphasizes grouping coun-
tries - regionalization is a subcase - in order to
pursue commonalities more aggressively than
would be possible with pure country-by-country
adaptation.
Developing a contingency theory of choice that
operates this level of disaggregation would seem to
be a high priority.
An additional assumption that is worth discuss-
ing in this context is the textbook distinction
between horizontal MNEs that emphasize aggrega-
tion and vertical MNEs that emphasize arbitrage.
This dichotomy assumes that it is often possible -
and useful - to distinguish firms in terms of the one
function that is economically central, over long
periods of time, to their strategies for adding value
by competing around the world.l1 If one accepts
this, then it is clear that there are two mutually
exclusive approaches to achieving geographic
coherence or fit - the international business
analogue of mainstream business strategy's focus
on internal and external fit at the level of the
individual business, and corporate strategy's focus
on fit or coherence across businesses. But there also
seem to be indications that large multinationals
engage, at least to some extent, in both aggregation
Journal of International Business Studies
 Semialobalization
-
and international
..----
business
150
and arbitrage. This naturally raises the question of
the extent to which it is possible to mix and match
across aggregation-oriented and arbitrage-oriented
activities. Or to put matters more starkly, how
feasible are transformation strategies that exten-
sively exploit both aggregation opportunities and
arbitragepossibilities?
A final question concerns whether intra-firm
cross-bordereconomic activity should be seen as a
substitute for or driver of market integration. It is
customary to think of (cross-border) firms as
remedies for the infirmities of (cross-border) mar-
kets. However, the importance of intra-firm trade
and FDI, in particular, hints that it might make
sense to shift towards seeing firms as global
connectors or conduits responsible, to a significant
extent, for cross-border integration rather than as
islands embedded in seas of market relationships.
Of course, whether firms' cross-border activities
substitute for or complement the cross-border
integration of markets is yet another open and
obviously important agenda item for future
research.
Conclusions
Accounts of the cross-borderintegration of markets
have tended to get very wrapped up in the times in
which they were written - perhaps too much so.
Thus Deutsch and Eckstein (1961) emphasized that,
by the 1950s, the internationalization of transac-
tions had declined significantly since the beginning
of the 20th century, and averredthat this trend was
unlikely to be reversed any time soon. Contrary to
their predictions, cross-border economic activity
surged in the 1960s onward and, as it breached
prewar records, inspired forked responses. Global-
ists stressed that international economic integra-
tion had reached new heights, while skeptics
insisted that it had barely returned to levels
experienced nearly a century earlier. Globalists
gained confidence with the fall of the Berlin Wall
in 1989 and the rapid growth in much of Asia
through much of the 1990s. But then came the
Asian financial crisis, episodes of instability in
Russiaand Latin America, a perceived 'globalization
backlash,' a global economic slowdown, and the
war on global terrorism. By mid-2002, the mood, at
least among practitioners, seemed to be one of
skepticism rather than optimism about globaliza-
tion.
The empirical evidence reviewed in this article
suggests that it might be preferable to take a more
measured, historically self-conscious perspective on
Journal of International Business Studies
strateav
--- -'
Pankai Ghemawat
cross-border integration instead of frequently
announcing changes in its direction or speed.
Specifically, the empirical review indicated that
most measures of market integration have scaled
new heights in the last few decades, but still fall far
short of economic theory's ideal of perfect integra-
tion. Looking forward, levels of cross-border inte-
gration may increase, stagnate or even suffer a
sharp reversalif the experience between and during
the two World Wars is any indication of the
possibilities: while technological changes may be
irreversible, political changes need not be. But
given the parameters of the current situation, it
seems unlikely that increases will any time soon
yield a state in which the differences among
countries can be ignored, multinationals' best
efforts to connect markets across borders notwith-
standing. Or that decreases could soon lead to a
state in which cross-border linkages can be for-
gotten about. So, one does not have to make a
precise forecast to diagnose that semiglobalization
as a condition is sufficiently broad to persist for
some time to come. Achieving similar stability in
attitudes toward cross-border operations would
seem preferable to manic-depressive swings in
attitudes about the outlook, if only for purely
pragmatic reasons.
The diagnosis of semiglobalization does more
than just supply a relatively stable frame of
reference for thinking about the environment of
cross-border operations. Semiglobalization also
calls attention to the critical role that location-
specificity plays in the prospects of distinctive
content for international business strategy relative
to mainstream business and corporate strategy. In
addition, it flags factors/products subject to loca-
tion-specificity as being salient from the perspec-
tive of international business. And, finally, it
highlights the scope for strategies that strive to
capitalize on the (large) residual barriers to cross-
border integration, as well as those that simply try
to cope with them.
Such considerations motivate the modest
proposal that semiglobalization or location-
specificity merits the status of a major research
program in international business. In other words,
that a significant volume of researchactivity should
be redirected along lines that take explicit account
of both the importance and the incompleteness of
the integration of markets across borders. In
addition to reflecting empirical reality, a research
program of this sort would directly address the
apparent dearth of 'big research questions' in
 Semiqlobalization and international business strateqy
international business. As Buckley (2002, 370)
recently put it:
International business has succeeded because it has focused
on, in sequence, a number of big questions, which arise
from empirical developments in the world economy. The
agenda is stalled because no such big question has currently
been identified. This calls into question the separate
existence of the subject area. It raises the old problem of
the relationship between international business and other
functional areas of management and social science.
From this perspective, the issue is not whether a
big research question is needed at this juncture in
the development of international business, but,
instead, what it should be about: semiglobalization /
location-specificity or something else?
Acknowledgements
This paper has benefited from research assistance by
Jamie Matthews and Raluca Lupu, helpful comments
by David Collis, Beulah D'Souza, Vijay Govindarajan,
Mauro Guillen, Tarun Khanna, Walter Kuemmerle,
Christos Pitelis, Ravi Ramamurti, Louis T. Wells Jr.,
George S. Yip and, especially, Bernard Y. Yeung, as
well as from presentation of material at the AIB Panel
Session, in summer 2002, celebrating Buckley and
Casson (1976). The Division of Research at the Harvard
Business School provided financial support.
Notes
'While location-specificity can also operate at the
local or (intranational) regional level, a full treatment
of it at all these levels of analyses is beyond the scope
of this paper, even though many of the analytical
issues that arise are similar.
2The disparity is even greater if one recognizes that
the denominator of the ratio should really be a
measure of gross sales rather than a value-added
measure like GDP.
3Foreign direct investment currently accounts for
roughly one-half of total foreign investment, but its
share was significantly smaller at the start of the 20th
century. See Bloomfield (1968, 3-4), cited in Bordo
et al. (1999).
4Note that the spread of domestic safety nets does
increase the likelihood that banking crises will turn into
currency crises.
References
Alcacer, j. and Chung, W. (2001) 'Knowledge seeking, human
capital and location choice of foreign entrants in the United
States', Working Paper, Stern School of Business, New York
University.
PankajGhemawat
151
5For further discussion of currency risk, see Frankel
(1992).
6Note the caveat that the extent of catch-up by the
Asian tigers would look somewhat less remarkable if
the data in Figure 5 were updated to take account of
the Asian currency crisis.
7The other (overlapping) reasons for the localization
of international competitiveness identified by Porter
(1990) are sophisticated local demand and the local
availability of specialized inputs and complements as
well as basic factors of production.
8This point can be demonstrated formally in the
context of standard supply-demand analysis. To start at
one extreme, with complete insulation between two
country markets, the price and quantity outcomes can
be pinned down (under the assumption of atomistic
competition) at the intersection of supply and demand
curves in each market. At the other extreme, with
complete integration - that is, zero extra costs of
trading, transporting, transacting and so on across
national boundaries - one could still add up the supply
curves for the two markets on the one hand and their
demand curves on the other and use the point of
intersection of the two aggregate curves to determine
the (common) prices and the quantities in the unified
market. But the continuum of situations between zero
and complete economic integration that I refer to as
semi-globalization creates additional challenges. Given
semi-globalization, the analysis of prices and quantities
in the two markets cannot be reduced to supply-
demand analysis of an individual market. Instead,
attention has to be paid to distinct markets that are
neither totally segmented nor totally integrated - an
intrinsicallymore complex, and interesting, setup.
9For further discussion along these lines, see Caves
(1998).
'?Such integration would make access to a global
pool of capital a 'given' for any worthy enterprise and
thereby limit the scope for purely financial sources of
advantage or disadvantage.
"Caves (1996) also identifies a third, residual
category of multinational enterprise: international
diversifiers whose operations in different countries
are neither horizontally nor vertically related to each
other. These can be thought of as falling in domain 4
of Table 5 rather than domain 3.
Baldwin, R.E. and Martin, P. (1999) Two Waves of Globalization:
Superficial Similarities, Fundamental Differences, National
Bureau of Economic Research: Cambridge, MA, NBERWorking
Paper No. 6904.
Journal of International Business Studies
 - - --=1 and international business strateav
Semiglobalization----
- - -- - -- ------
152
Barro, R.J. and Sala-i-Martin, X. (1995) Economic Growth,
McGraw-Hill: New York.
Bekaert, G. and Harvey, C.R. (1995) 'Time-varying world market
integration', Journal of Finance 50(2): 403-444.
Blonigen, B.A. (1997) 'Firm-specific assets and the link between
exchange rates and foreign direct investment', American
Economic Review 87(3): 447-465.
Bloomfield, A.I. (1968) Patterns of Fluctuationin International
Finance Before 1914, Princeton Studies in International Finance
No. 21, International Finance Section, Department of Eco-
nomics, Princeton University.
Bordo, M.D., Eichengreen, B. and Irwin, D.A. (1999) Is
Globalization Today Really Different than Globalization a
Hundred Years Ago?, National Bureau of Economic Research:
Cambridge, MA, NBERWorking Paper 7195 (Prepared for the
Brookings Trade Policy Forum on Governing in a Global
Economy, Washington, DC, 15-16 April 1999. Published:
shorter version in Austrian Economic Papers, Vols. 1 and 2,
2000).
Buckley, P.J. (2002) 'Is the international business agenda
running out of steam', Journalof InternationalBusinessStudies
33(2): 365-373.
Buckley, P.J. and Casson, M. (1976) The Futureof the Multi-
national Enterprise,Macmillan:London.
Caves, R.E.(1996) MultinationalEnterpriseand EconomicAnalysis,
Cambridge UniversityPress:Cambridge.
Caves, R.E.(1998) 'Researchon internationalbusiness: problems
and prospects', Journalof InternationalBusinessStudies29(1):
5-19.
Coase, R.H. (1937) 'The nature of the firm', Economica4(16):
396-405.
Deutsch, K.W.and Eckstein,A. (1961) 'National industrialization
and the declining share of the internationaleconomic sector,
1890-1959', WorldPolitics13(2): 267-272.
Dune, N. (2001) 'US population is now more than 10%
foreign-born', FinancialTimes,4 January2001, The Americas,
p. 4.
Feenstra, RC (1998) 'Integration of trade and disintegration of
production in the global economy', Journal of Economic
Perspectives12(4): 31-50.
Feldstein, M (1995) 'The Effects of Outbound Foreign Direct
Investment on the Domestic Capital Stock', in M. Feldstein, J.
Hines and R.G. Hubbard (eds.) The Effects of Taxation on
MultinationalCorporations,Universityof Chicago Press: Chi-
cago, pp: 43-63.
Feldstein, M. and Horioka, C. (1980) 'Domestic savings and
international capital flows', EconomicJournal90(358): 314-
329.
Frankel,J.A.(1992) 'Measuring internationalcapital mobility: a
review', AmericanEconomicReview82(2): 197-202.
Frankel,Jeffrey,A. (2001) 'Assessing the EfficiencyGain from
FurtherLiberalization',in Porter,Roger B., PierreSauve, Arvind
Subramanian & Americo Beviglia Zampetti, (eds.) Efficiency,
equity, and legitimacy: the multilateraltrading system at the
millennium,BrookingsInstitution Press:Washington, D.C.
Ghemawat, P. (2000) 'Global advantage: arbitrage, replication,
and transformation', unpublished note, Harvard Business
School, December.
Giddens, A. (1996) 'Keynote address at the United Nations
Research Institute for Social Development, as excerpted in
'Essentialmatter", UNRISDNews, No. 15.
Acceptedby Tom Brewer;outgoingEditor,August 2002.
Journal of International Business Studies
-- - - -- -
----f
Pankai Ghemawat
- "-)-
-- - ,-
Held, D., McGrew, A.G., Goldblatt, D. and Perraton,J. (1999)
Global Transformations:Politics,Economics,and Culture,Stan-
ford UniversityPress:Stanford.
Helliwell, J.F. (1998) How Much Do National BordersMatter?,
BrookingsInstitutionPress:Washington, DC.
Irwin,D.A. (1996) 'The United States in a new global economy?
A century's perspective', AmericanEconomicReview86(2): 41-
46.
Jaffe,A.B. and Trajtenberg,M. (1999) 'Internationalknowledge
flows: evidence from patent citations', Economicsof Innovation
and New Technology8(1-2): 105-1 36.
Keller, W. (2000) Geographic Localization of International
TechnologyDiffusion,National Bureau of Economic Research:
Cambridge, MA, NBERWorking Paper No. 7509.
Kenwood, A.G. and Lougheed, A.L. (1989) The Growthof the
InternationalEconomy,1920-1960, Allen & Unwin: London.
Lewis, K.K.(1995) 'Puzzles in InternationalFinancialMarkets',in
G. Grossman and K. Rogoff (eds.) Handbookof International
Economics,Vol. III, Elsevier Science: Amsterdam. pp: 1913-
1971.
Maddison, A. (1995) Monitoringthe WorldEconomy:1820-1992,
Development Centre of the Organization for Economic
Cooperation and Development: Paris.
Martin, S.F. (ed.) (2000) World Migration Report: 2000,
Copublished by the InternationalOrganization for Migration
and the United Nations.
McCallum, J. (1995) 'National borders matter: Canada-US
regional trade patterns', American Economic Review 85(3):
615-623.
Micklethwait,J. and Wooldridge, A. (2000) A FuturePerfect:The
Challenge and Hidden Promise of Globalization,Crown Busi-
ness: New York.
Obstfeld, M. and Taylor, A. (1997) The Great Depressionas a
Watershed:InternationalCapital Mobilityover the Long Run,
National Bureauof Economic Research:Cambridge, MA, NBER
Working Paper 5960.
Patel, P. and Pavitt, K. (1994) National InnovationSystems:Why
they are Important and How they Might be Measured and
Compared,Mimeo, Science Policy ResearchUnit, Universityof
Sussex.
Penrose, E.T.(1959) The Theoryof the Growthof the Firm,Basil
Blackwell,Wiley: Oxford, New York.
Porter, M.E. (1990) The CompetitiveAdvantage of Nations, Free
Press: New York.
Prahalad, C.K. and Doz, Y. (1987) The MultinationalMission:
Balancing Local Demands and Global Vision,Free Press: New
York.
Rogoff, K. (1996) 'The purchasing power parity puzzle', Journal
of EconomicLiterature34(2): 647-668.
Scott, B. (2000) EconomicStrategiesof Nations, Mimeo, Harvard
Business School.
Teece, D.J. (1977) 'Technology transfer by multinationalfirms:
the resourcecost of transferringtechnological know-how', The
EconomicJournal87(346): 242-261.
UN Center on Transnational Corporations, World Investment
Report(various issues).
World Economic Forum,Global Competitiveness Report(various
issues).